The tech sell-off is no longer a vague warning about rich AI valuations. SpaceX, Samsung, SK Hynix and the Nasdaq are now showing you what happens when investors ask for cash returns instead of promises.
The question was never whether a tech correction would arrive. It was what would finally make investors stop treating every AI infrastructure story as if the ending had already been written. On Monday, June 22, and into Tuesday, June 23, the answer was blunt: debt, stretched valuations and a sudden refusal to keep paying any price for future compute demand.
As Business Insider reported, SpaceX shares were set to open around $150 on Tuesday after touching an intraday high of $225 the previous week. That is still above the $135 IPO price, but it is below the $160.95 closing price from its first trading day on June 12, according to MarketWatch. Anyone who bought after that first-day close is now underwater, at least on paper. For a company that just completed the largest IPO in history, that is a fast lesson in how little patience public markets have once the first rush fades.
The damage was not small. MarketWatch, citing Dow Jones Market Data, put SpaceX's Monday close at $154.60, with a market value of about $2.04 trillion. Business Insider said the drop had erased about $400 billion from the company's closing market value after its June 16 peak. The stock was still enormous, but the air had clearly come out of the first trade.
Look at the trigger and you can see why the selling spread. The New York Post reported that SpaceX disclosed it may raise around $20 billion through investment-grade bonds, even after holding about $100.8 billion in cash. Investors can tolerate ambition. They get nervous when ambition arrives with a financing plan attached to it, especially when the plan is tied to AI projects that still have to prove they can earn back the money being poured into them.
The chip sector took the next hit. Barron's reported that Micron fell about 7% in pre-market trading Tuesday after Samsung Electronics and SK Hynix both dropped more than 12% in Seoul. South Korea's Kospi fell nearly 10%, triggering a 20-minute trading halt, according to MarketWatch. That is not just a bad morning for a few crowded trades. Samsung and SK Hynix sit right inside the memory supply chain feeding AI data centers, so when they fall together, you should not pretend the scare is local.
There is a reason this landed harder than the usual Nasdaq wobble. AI spending has been treated as a kind of self-validating loop: data centers need chips, chips need memory, memory makers book demand, cloud companies promise future AI revenue, and the market rewards the whole chain before the end customer has paid enough to settle the argument. That works until investors ask a basic question. Where is the profit?
The market wants proof now
For two years, companies across the technology sector have announced huge capital expenditure plans, signed chip orders, leased power, built data centers and asked investors to wait for the applications that will make all of it pay. That was acceptable while rates, momentum and earnings revisions were all moving in the right direction. It gets harder when a newly public SpaceX is talking about more debt and South Korea's two most important chip stocks are falling by double digits in one session.
CNBC reported that S&P 500 futures fell 1.45% and Nasdaq 100 contracts dropped 2.69% in early Tuesday trading, while the iShares Semiconductor ETF was down 6.2% before the open. Intel, Micron and AMD were all lower in pre-market trade. Those are the numbers you watch if you're trying to understand whether this is one company having a post-IPO hangover or a wider repricing of AI risk.
Frankly, a repricing was overdue. The market had started to treat capacity as revenue and revenue as profit before the chain had been tested. SpaceX is the cleanest example because the stock moved so violently, but it is not the only one. If a company can lose hundreds of billions of dollars in market value within days and still be worth around $2 trillion, the problem was never a lack of investor enthusiasm. The problem was that enthusiasm had become the valuation method.
The international piece matters too. When US mega-cap tech sells off, you can tell yourself the issue is a handful of expensive names. When Samsung, SK Hynix and the Kospi are dragged into the same trade, the story is bigger. AI infrastructure is not built in one country, and neither is the risk. Memory chips, advanced packaging, cloud spending, power contracts, bond markets, you name it, they all meet inside the same assumption that demand will keep rising fast enough to justify the bill.
There is still a version of this where earnings calm the market. Strong cloud revenue, clear AI product margins and steadier guidance could do some real work. But investors are no longer buying the whole story on faith. SpaceX's slide below its first-day closing price, the Kospi's 10% fall and the double-digit losses at Samsung and SK Hynix have made the point plainly enough. The AI boom may be real. The prices attached to it were never guaranteed.
Also read: Abu Dhabi's MGX has quietly become the most consequential AI investor on the planet • Google lost two of its most important AI researchers within days and the market wiped out $250 billion to make the point • Bending Spoons is heading to Nasdaq at a $19 billion valuation and its turnaround math is hard to argue with
